Evolutionary Applications. 2018;1–21. wileyonlinelibrary.com/journal/eva
|
1 Received: 15 March 2018|
Revised: 19 May 2018|
Accepted: 22 May 2018DOI: 10.1111/eva.12652 O R I G I N A L A R T I C L E
The ghost of introduction past: Spatial and temporal variability
in the genetic diversity of invasive smallmouth bass
Genevieve Diedericks
1,2| Romina Henriques
3| Sophie von der Heyden
2|
Olaf L. F. Weyl
4,5| Cang Hui
6,7This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2018 The Authors. Evolutionary Applications published by John Wiley & Sons Ltd 1Centre for Invasion Biology, Department of Botany and Zoology, Stellenbosch University, Matieland, Stellenbosch, South Africa 2Evolutionary Genomics Group, Department of Botany and Zoology, Stellenbosch University, Matieland, Stellenbosch, South Africa 3Section for Marine Living Resources, National Institute of Aquatic Resources, Technical University of Denmark, Lyngby, Denmark 4DST/NRF Research Chair in Inland Fisheries and Freshwater Ecology, South African Institute for Aquatic Biodiversity (SAIAB), Grahamstown, South Africa 5Centre for Invasion Biology, South African Institute for Aquatic Biodiversity (SAIAB), Grahamstown, South Africa 6Centre for Invasion Biology, Department of Mathematical Sciences, Stellenbosch University, Matieland, Stellenbosch, South Africa 7Mathematical Biosciences Group, African Institute for Mathematical Sciences, Cape Town, South Africa Correspondence Genevieve Diedericks, Centre for Invasion Biology, Department of Botany and Zoology, Stellenbosch University, Private Bag X1, Matieland, 7602, Stellenbosch, South Africa. Email: diedericks.genevieve@gmail.com Funding information Department of Science and Technology, Republic of South Africa, Grant/Award Number: 109015 and 110507; South African Agency for Science and Technology Advancement, Grant/Award Number: 109244 and 89967; DST-NRF Centre of Excellence for Invasion Biology
Abstract
Understanding the demographic history of introduced populations is essential for unravelling their invasive potential and adaptability to a novel environment. To this end, levels of genetic diversity within the native and invasive range of a species are often compared. Most studies, however, focus solely on contemporary samples, rely-ing heavily on the premise that the historic population structure within the native range has been maintained over time. Here, we assess this assumption by conducting a three- way comparison of the genetic diversity of native (historic and contempo-rary) and invasive (contempoa three- way comparison of the genetic diversity of native (historic and contempo-rary) smallmouth bass (Micropterus dolomieu) popula-tions. Analyses of a total of 572 M. dolomieu samples, representing the contemporary invasive South African range, contemporary and historical native USA range (dating back to the 1930s when these fish were first introduced into South Africa), revealed that the historical native range had higher genetic diversity levels when compared to both contemporary native and invasive ranges. These results suggest that both con-temporary populations experienced a recent genetic bottleneck. Furthermore, the invasive range displayed significant population structure, whereas both historical and contemporary native US populations revealed higher levels of admixture. Comparison of contemporary and historical samples showed both a historic introduction of M. do-lomieu and a more recent introduction, thereby demonstrating that undocumented introductions of this species have occurred. Although multiple introductions might have contributed to the high levels of genetic diversity in the invaded range, we dis-cuss alternative factors that may have been responsible for the elevated levels of genetic diversity and highlight the importance of incorporating historic specimens into demographic analyses. K E Y W O R D S demographic history, genetic bottleneck, genetic diversity, historic DNA, invasive, multiple introductions, sampling design
1 | INTRODUCTION
Understanding the demographic history of populations consti- tutes a fundamental aspect of evolutionary biology. Invasive spe-cies are particularly suitable for demographic analyses, as they frequently experience rapid alternations in levels of genetic diver-sity following introduction (Chown et al., 2015; Hui & Richardson, 2017; Lee, 2002; Rius & Darling, 2014; Roman & Darling, 2007). To this end, the assessment of genetic diversity has become es-sential for establishing the demographic and adaptive potential of populations in novel environments (Dlugosch, Anderson, Braasch, Cang, & Gillette, 2015; Prentis, Wilson, Dormontt, Richardson, & Lowe, 2008; Stapley, Santure, & Dennis, 2015; Zenni, Bailey, & Simberloff, 2014) and provides insight into the role that genetic variability plays in an organisms’ invasive success (Edelaar et al., 2015). Ultimately, this information allows predictions to be made on population viability, aiding in the development of an appropri-ate, species- specific management strategy (Chown et al., 2015; Meyer et al., 2017; Prentis et al., 2008).
Numerous studies have attempted to assess the effects of in-vasion dynamics on genetic variation (e.g., founder effects, genetic bottlenecks, admixture, propagule pressure; Baker & Stebbins, 1965; Hui & Richardson, 2017; Mayr, 1963) by comparing populations in the native and invasive ranges (Kelly, Muirhead, Heath, & Macisaac, 2006; Kolbe et al., 2004; Naccarato, Dejarnette, & Allman, 2015; Rollins, Woolnough, Wilton, Sinclair, & Sherwin, 2009). These types of studies aid in unravelling the demographic history of the invasive species in question (Ficetola, Bonin, & Miaud, 2008; Gillis, Walters, Fernandes, & Hoffman, 2009; Gray et al., 2014; Neilson & Stepien, 2011). Yet, despite the wealth of specimens and information housed within Natural History collections, the majority of invasion studies to date have focussed exclusively on contemporary populations, thereby relying heavily on the premise that the historic population structure within the native range has been maintained over time.
Historic DNA serves as a valuable reference when examining contemporary genetic diversity (Bouzat, 2000; Dormontt et al., 2014; Guinand, Scribner, & Page, 2003; Lozier & Cameron, 2009), as it allows for the monitoring of temporal changes in genetic diversity across generations (Guinand et al., 2003; Sefc, Payne, & Sorenson, 2007). This temporal approach increases the chance of detecting subtle changes frequently overlooked by studies focussing only on contemporary data (Lozier & Cameron, 2009) and thus allows us to delineate the most likely invasion scenario (Gillis et al., 2009; Thompson et al., 2011; Van Kleunen, Weber, & Fischer, 2010) and reveal connectivity levels among invasive populations (Beneteau, Walter, Mandrak, & Heath, 2012; Funk, Garcia, Cortina, & Hill, 2011; Snyder & Stepien, 2017). This may be of particular importance in studies conducted on taxa for which there is a priori reason to sus-pect temporal fluctuations in genetic variation, such as highly ex-ploited (and subsequently stocked) taxa or species often associated with human- mediated dispersal. Hence, from an evolutionary per-spective, the incorporation of historic DNA is therefore of funda-mental importance.
Smallmouth bass, Micropterus dolomieu (Lacepèdé, 1802), pres-ents a suitable model system to investigate variation in genetic diversity through space and time, as the species’ exploitation and subsequent stocking events within the native range are well doc-umented (Long, Allen, Porak, & Suski, 2015), and its formal intro-duction history and subsequent spread into and throughout South Africa are well recorded (De Moor & Bruton, 1988). Twenty- nine
M. dolomieu specimens originating from broodstock collected in
the Wheeling River, West Virginia, USA, were shipped from the Lewistown hatchery in Maryland, USA, to the Jonkershoek hatch-ery in South Africa in 1937 (De Moor & Bruton, 1988; Loppnow, Vascotto, & Venturelli, 2013; Powell, 1967). Here, they were reared and bred before being released into multiple water bodies across the country to provide opportunities for angling (De Moor & Bruton, 1988). Most of the documented stockings (De Moor & Bruton, 1988) occurred prior to the cessation of government support to stocking programs in the early 1990s (Ellender, Woodford, Weyl, & Cowx, 2014).
Considering that both the historical record and contemporary distributions of M. dolomieu in South Africa are well documented, this study aims to (a) assess the genetic differentiation and diver-sity within M. dolomieu populations in South Africa, (b) investigate how genetic diversity changed over time in both native and invasive ranges, and (c) assess the introduction history of M. dolomieu into South Africa. Given the small M. dolomieu founding population, we predict that the invasive South African range will have a lower ge- netic diversity when compared to the native (historic and contem-porary) North American range due to a loss of alleles, as suggested by Dlugosch and Parker (2008). Furthermore, as heavily exploited species often experience genetic bottlenecks, leaving traces in the species’ genetic diversity (Pinsky & Palumbi, 2014), we predict that the genetic diversity will be lower in contemporary time when com-pared to historical samples in the native range.
2 | MATERIALS AND METHODS
2.1 | DNA collection and extraction from historical
specimens
Specimens representing the historical native range (Figure 1), cor-responding to the approximate time of introduction into South Africa (1930–1941), were obtained from a host of collections housed at the Smithsonian National Museum of Natural History (NMNH), The Academy of Natural Sciences of Drexel University (ANSP), University of Michigan Museum of Zoology (UMMZ) and the Ohio State University Museum (OSUM) (Table 1; Appendix 1). In total, 53 formalin- fixed specimens representing 11 drainage systems were obtained for genetic analyses (Table 1). These specimens represent a subset of the M. dolomieu genetic diversity that was present in the native range 20–25 generations ago (Barthel et al., 2008).
Genomic DNA was extracted from preserved muscle tissue (20–50 mg) in a room previously unexposed to fish DNA using steril-ized equipment. Prior to each extraction, all equipment and surfaces
were treated with 10% bleach to remove any potential contaminants. Pikor, Enfield, Cameron, and Lam (2011) showed that high- quality DNA can be extracted from formalin- fixed tissue if the samples are rehydrated with a series of ethanol washes prior to extraction. Thus, 500 μl of 100% ethanol was added to each tissue sample and vor-texed vigorously for 30 s. The liquid was removed, and the process was repeated with 500 μl 70% ethanol, followed by 1,000 μl distilled water. Lastly, 1,000 μl distilled water was added to each sample and left to soak at 55°C for 5 days, vortexing the sample every 24 hr. Once rehydrated, the sample was moved to a dry Eppendorf tube be-fore DNA extraction, using the QIAamp DNA FFPE tissue extraction kit (QIAGEN). In a recent review, Paireder et al. (2013) demonstrated that this kit consistently outcompeted other extraction methods when working with old (1820–1950), formalin- fixed tissue. Apart from doubling the amount of proteinase K added to each sample (60 μl), extraction followed the manufacturers’ protocol. To break the formalin bonds, the samples were heated to 90°C for 1 hr before commencing with the wash steps. Lastly, to ensure the maximum elution of bound DNA, 10 μl elution buffer (warmed to 25.5°C) was added and left to “incubate” at room temperature for 5 min before centrifuging at 20,000 g for 1.5 min. This was repeated three times to yield a total DNA extraction volume of 30 μl. All DNA extractions were stored at −20°C.
2.2 | DNA collection and extraction from
contemporary specimens
Fresh tissue samples (muscle, liver, fin clippings) were derived from specimens collected by angling in both the native United States of America (USA) and Canada and the invasive South African (SA) ranges during the summer months of 2014 and 2015 (Figure 1). Collections in North America were led by a host of individuals and organizations based in the USA and Canada (see Acknowledgements). Nine localities rendering a total of 213 specimens were sampled
from the same “broad” area represented by the historical samples to allow for direct genetic diversity comparisons (Table 1). Additional specimens collected in 2014 (n = 7; formalin fixed), representing the Detroit River, were obtained from the Royal Ontario Museum (ROM), Canada.
All SA specimens were euthanized with clove oil (CapeNature permit number 0056- AAA043- 00004; Eastern Cape permit num-bers CRO 165/14CR and CRO 166/14CR; Mpumalanga permit number MPB. 5498/2; Ethical clearance reference number SU- ACUM14- 00011, University of Stellenbosch) before sampling a piece of tissue. Tissue samples were stored in 70% ethanol for subse-quent DNA extraction. Additional specimens (n = 63) were obtained from the South African Institute for Aquatic Biodiversity (SAIAB), Grahamstown, South Africa, rendering a total sample size of 306 spec-imens representing eight river systems (Table 1; Appendix 1). DNA was extracted from each contemporary specimen (USA & SA) using the NucleoSpin Tissue extraction (gDNA) kit (MACHEREY- NAGEL, Separations, Cape Town, South Africa) following the manufacturers’ protocol. All DNA extractions were stored at −20°C.
2.3 | Historical and contemporary DNA
amplification
To corroborate the morphological identification of the contem- porary collected specimens and assess genetic diversity and de-mographic history of both native and invasive populations, two partial mitochondrial DNA (mtDNA) gene regions, namely cy-tochrome b (cytb) and control region (CR), were amplified for all the contemporary samples (n = 519). This was not possible for the historical samples due to the limited availability of tissue and the degraded nature of the DNA. A standard 25 μl master-mix was prepared for both mtDNA polymerase chain reactions (PCRs). The internal cytb primers, basscytbf1 (5′- CAC CCC TAC TTC TCC TAC AAA GA- 3′) and basscytbr1 (5′- AAG GCR AAG
F I G U R E 1 Map of native USA (left) and invasive SA (right) sampling localities. Letters A- K denote historical sampling localities, while
numbers denote contemporary sampling localities. All letters and numbers correspond to those used in Table 1. The location indicated by the star (i) represent the Wheeling River, while the downward- facing arrows denote the (ii) Lewistown hatchery and (iii) Jonkershoek hatchery, respectively
T A B LE 1 A n ov er vi ew o f t he s am pl ed p op ul at io ns fr om th e co nt em po ra ry inv as iv e (C I), c on te m po ra ry n at iv e (C N ) a nd h is to ric al n at iv e (H N ) r an ge s. A bb re vi at io ns c or re sp on d to th os e us ed in s ub se qu en t t ab le s, te xt a nd A pp en di x1 N at ive/ in va si ve St at e/ pr ov in ce C ol le ct ion da te Sa m pled loc al it y D ra in ag e sy st em A bb r. i n Tab les N For ma lde hy de ex pos ur e Ma te ria l su ppl ied B y Sy mbol on s am pl in g m ap (Fi gu re 1 ) H is to ric al sp eci m ens N at iv e O hio 19 30 W hi te O ak C re ek O hi o R iv er OH 3 Ye s O SU M A N at iv e O hio 19 40 ; 19 41 Au gla iz e Ri ve r Au gla iz e Ri ve r AU 5 Ye s O SU M B N at iv e Mich ig an ; O nt ar io 19 34 ; 1 93 5; 1 94 0 D et ro it Ri ve r D et ro it Ri ve r DE T 18 Ye s U M M Z C N at iv e O hio 19 41 La ke E rie La ke E rie LE 3 Ye s O SU M D N at iv e O hio 19 38 M os qu ito C re ek L ak e Mo sq ui to Cr eek MO 2 Ye s O SU M E N at iv e N ew Y or k 19 37 A lle gh en y Riv er A lle gh en y Riv er AL 3 Ye s U M M Z F N at iv e N ew Y or k 19 31 Fa ll C re ek C ay ug a La ke , E tn a FC 2 Ye s U M M Z G N at iv e N ew Y or k 19 35 O ts el ic R iv er ; Su sq ueh an na R iv er Su sq ueh an na R iv er SU 5 Ye s U M M Z H N at iv e N ew Y or k 19 36 Ro ndo ut R iv er H ud so n Riv er HUD 4 Ye s U M M Z I N at iv e M ar yl an d 19 41 Mo no cac y R iv er Po to m ac R iv er PO 4 No A NS P J N at iv e V irg in ia ; W es t V irg in ia 19 33 –19 36 Sh en an doa h Riv er Sh en an doa h Riv er SH 4 Ye s NMNH K 53 C on te mp or ar y Sp eci m ens N at iv e O nt ar io 201 3; 2 01 4 D et ro it Ri ve r D et ro it Ri ve r DE T 7 Ye s RO M 1 N at iv e N ew Y or k 20 14 N ia gr a Ri ve r N ia gr a Ri ve r NI A 49 No U SA c ol le ct or s 2 N at iv e N ew Y or k 20 14 St L aw re nc e Ri ve r St L aw re nc e Ri ve r STL 55 No U SA c ol le ct or s 3 N at iv e N ew Y or k 201 5 O ne id a La ke O neid a R iv er O N EI 27 No U SA c ol le ct or s 4 N at iv e N ew Y or k 201 5 Sa ra to ga L ak e H ud so n Riv er SA R 10 No U SA c ol le ct or s 5 N at iv e N ew Y or k 201 5 Ve st al ; S usq ueh an na R iv er Su sq ueh an na R iv er VE S 14 No U SA c ol le ct or s 6 N at iv e N ew Y or k 201 5 O ne on ta ; S us qu eh ann a Riv er Su sq ueh an na R iv er O N EO 10 No U SA c ol le ct or s 7 N at iv e N ew Y or k 201 5 Lo lli er sv ill e Su sq ueh an na R iv er LO L 20 No U SA c ol le ct or s 8 N at iv e N ew Y or k 20 14 H ud so n Riv er H ud so n Riv er HUD 21 No U SA c ol le ct or s 9 213 In va si ve W es te rn C ap e 20 14 D or in g Ri ve r D or in g Ri ve r DO 38 No Se lf- col le ct ed 1 In va si ve W es te rn C ap e 20 14 O lif an ts R iv er ; J an D is se ls Riv er O lif an ts R iv er OL 44 No Se lf- col le ct ed 2 In va si ve W es te rn C ap e 20 14 B er g Riv er B er g Riv er BE 22 No Se lf- col le ct ed 3 In va si ve W es te rn C ap e 20 14 B re ede R iv er B re ede R iv er BR 43 No Se lf- col le ct ed 4 In va si ve Ea st er n C ap e 20 14 K ou ga R iv er K ou ga R iv er KO 46 No Se lf- col le ct ed 5 In va si ve Ea st er n C ap e 201 2 K ro m R iv er K ro m R iv er KR 15 No SAIAB 6 In va si ve Ea st er n C ap e 20 14 Ro oi kr an z D am B uf fa lo R iv er B U 48 No SAIAB 7 In va si ve M pu mal ang a 20 14 B ly de D am B ly de R iv er M P 50 No M PB 8 306
CGG GTG AGG G- 3′; Near, Kassler, Koppelman, Dillman, & Philipp, 2003), were used to amplify the cytb fragment. The primer set CB3R- L (5′- CATATTAAACCCGAATGATATTT- 3′; Palumbi, 1996) and HN20- R (5′- GTGCTTATGCTTTAGTTAAGC- 3′; Bernatchez & Danzmann, 1993) was used to amplify the CR. Both PCR reactions followed the authors’ protocols. All PCR products were visualized through gel electrophoresis before being sequenced (ABI 3730 XL DNA Analyzer, Applied Biosystems, CAF, Stellenbosch, South Africa). Chromatographs were visually inspected and aligned in
Geneious® 10.0.2 (Biomatters, Auckland, New Zealand).
Fifteen microsatellite loci, designed for both species- and genus- level amplification, were selected from published literature (Supporting Information Table S1). Of these, only 11 loci (eight species- specific: Mdo3, Mdo4, Mdo5, Mdo7, Mdo8, Mdo9, Mdo10, Mdo11—Malloy, Den Bussche, Jr, Coughlin, & Echelle, 2000; and three genus- specific: Lma21—Colbourne, Neff, Wright, & Gross, 1996; Lma102, Lma117—Neff, Fu, & Gross, 1999) were successfully amplified. As Lma102 and Lma117 were not polymorphic for a sub-set of specimens, they were excluded; therefore, nine polymorphic loci were used in the present study (Supporting Information Table S1). Three multiplex reactions were performed using the KAPA2G™ Fast Multiplex PCR Kit (KapaBiosystems, Cape Town, South Africa). The same nine microsatellite loci were amplified for the historic samples, following the amplification procedure used for the contem-porary DNA, but due to the degraded nature of the DNA, this did not yield results. Thus, the resulting PCR products for each multiplex were diluted with distilled water to obtain a 1/10 PCR product which, in turn, served as template in the subsequent PCR. To ensure ampli-fication and to avoid the overestimation of genetic diversity often associated with the amplification of ancient- and formalin- fixed DNA (Buchan, Archie, Van Horn, Moss, & Alberts, 2005; Sefc et al., 2007), historical samples were amplified twice for each microsatellite locus. All microsatellite genotyping was performed on an ABI 3730 XL DNA Analyzer (Applied Biosystems, CAF, Stellenbosch, South Africa), using LIZ as an internal size marker, and scoring was conducted in Geneious® 10.0.2 (Biomatters, Auckland, New Zealand). To ensure accurate scoring, reference individuals previously scored were used as positive controls. Historical specimens were scored blindly (i.e., specimen name removed) and repeated three times to ensure accu-racy and consistency. Where scoring inconsistencies were observed (historical specimens) and more than three loci could not be scored (for both historical and contemporary specimens), the entire speci-men was removed from the data set and excluded from the study. Similarly, as one microsatellite locus, Mdo8, did not amplify for the majority of historical samples, it was removed from the historical data set entirely. Thus, nine microsatellite loci were analysed for the contemporary data set, but only eight microsatellite loci were anal-ysed for the historical data set.
2.4 | Contemporary mtDNA analyses
To assess genetic diversity levels in both the contemporary native (USA—CN) and invasive (SA—CI) ranges, the number of haplotypes
(H), haplotype diversity (h) and nucleotide diversity (π) were calculated for each sample site. The population history for M. dolomieu in both ranges was examined using Fu’s Fs (Fu, 1997) and Tajima’s D (Tajima, 1989). Assessment of genetic population structure was conducted combining both native and invasive contemporary data sets for each
gene fragment. Pairwise FST values were calculated and a hierarchical
analysis of molecular variance (AMOVA) conducted to determine the amount of population subdivision among sampled localities. All analyses were conducted in ARLEQUIN 3.5.2.2 (Excoffier & Lischer, 2010), with statistical significance assessed with 10,000 permutations.
2.5 | Contemporary and historical
microsatellite analyses
All microsatellite loci were assessed for linkage disequilibrium and deviations from Hardy–Weinberg equilibrium (HWE) in Genepop 4.2.1 (Rousset, 2008), with statistical significance assessed after 10,000 iterations. The Bonferroni method was used to correct for multiple comparisons (Rice, 1989). Amplification errors associated with large allele dropout and stuttering were assessed with MICROCHECKER 2.2.3 (Van Oosterhout, Weetman, & Hutchinson, 2006). As most of the populations were found to not comply with HWE assumptions, FreeNA 1.2 (Chapuis & Estoup, 2007) was used to check for the presence of null alleles using the EM algorithm (Dempster, Laird, & Rubin, 1977). Intraspecific and within- population genetic diversity levels were assessed as number of alleles (Na),
allelic richness (AR), observed (HO) and expected heterozygosity (HE), and Wright’s inbreeding coefficient (FIS), as implemented in FSTAT 2.9.3.2 (Goudet, 1995), Genepop 4.2 (Rousset, 2008), HP- Rare 1.1 (Kalinowski, 2005) and ARLEQUIN 3.5.2.2 (Excoffier & Lischer, 2010). Statistical significance of FIS was assessed after 1,000 permutations in FSTAT 2.9.3.2 (Goudet, 1995). Allelic richness (AR) was calculated using HP- Rare 1.1 (Kalinowski, 2005), correcting for sample size disparity through rarefaction analysis. Analyses were conducted per population for the two contemporary data sets, but due to the small sample size for most of the historical localities (Table 1), these were grouped (= MUS) to obtain the genetic diversity indices.
Multiple approaches were employed to investigate the popula-tion structuring and genetic connectivity among (contemporary and historical) populations. As only eight loci were successfully amplified for the historical native (HN) specimens, all comparative analyses in-corporating the historical samples only compared the eight loci, while contemporary SA–USA comparisons encompassed nine loci. First, to determine whether there was a difference in observed
hetero-zygosity (HO) between the three groups (CI, CN, HN), an analysis of
variance (ANOVA) was conducted in spssstatistics 20.0.0 (SPSS Inc.,
Chicago, IL, USA), with loci selected as random factors. Subsequently, a Bonferroni post hoc test was used to further assess the differences between groups. In addition, a stacked bar graph was constructed to visualize the variation among localities and loci. Second, Weir’s (1996) FST was employed to assess the genetic differentiation among sam-pled localities using FreeNA 1.2 (Chapuis & Estoup, 2007). FreeNA, employing the ENA correction method (Chapuis & Estoup, 2007), was
chosen as it has been shown to correctly estimate FST values in the presence of null alleles (detected in the previous analysis; Chapuis & Estoup, 2007). A jackknife approach with 1,000 bootstrap replicates was employed to assess statistical significance (Chapuis & Estoup, 2007). Next, BOTTLENECK 1.2.02 (Piry, Luikart, & Cornuet, 1999) was used to test the prediction that both contemporary populations (CI and CN) experienced a recent genetic bottleneck. Populations that have undergone a genetic bottleneck are often associated with a loss of (rare) alleles and display elevated levels of heterozygosity when compared to stable populations (Piry et al., 1999). Thus, signifi-cant heterozygote excess was evaluated for each of the three groups using a Wilcoxon rank test (10,000 iterations) for two mutational models often associated with microsatellite evolution: the two- phase mutation model (TPM) and the infinite alleles model (IAM). To investigate the genetic associations within each of the three groups as well as among them, without being influenced by the lack of HWE or the presence of null alleles, a principal component anal-ysis (PCA) using microsatellite allelic frequencies was conducted in the R package Adegenet 1.3.1 (Jombart & Ahmed, 2011). Next, we used STRUCTURE 2.3.4 (Pritchard, Stephens, & Donnelly, 2000) to (a) identify and visualize the population structure within each of the three groups (CI, CN and HN), (b) compare overlapping populations from the historical and contemporary native range and (c) search for a potential source population from where the invasive South African stocks originated. Four STRUCTURE analyses (each group independently followed by an analysis combining CI, CN and HN) were conducted using the admixture model with correlated allele frequencies, allowing each individual to be allocated to multiple clusters as determined by its genotype frequency. Five replicate runs were conducted for each K (1 < K < 15). Runs were conducted using an initial burn- in of 75,000 Markov chain Monte Carlo (MCMC) generations, followed by 350,000 MCMC steps. STRUCTURE HARVESTER 0.6.94 (Earl & vonHoldt, 2012) was used to deter-mine the most probable K following the Evanno method (Evanno, Regnaut, & Goudet, 2005), before using CLUMPP 1.1.2 (Jakobsson & Rosenberg, 2007) to compile the five replicate runs for the most likely K. DISTRUCT 1.1 (Rosenberg, 2004) was used to visualize the composite assignments.
At last, we performed an approximate Bayesian computation (ABC) on the microsatellite data set to determine whether the inva-sive South African M. dolomieu populations originated from a single introduction event from the USA as stated by the historical records, using DIYABC 2.1.0 (Cornuet et al., 2014). As null alleles were only observed in one locus (Mdo9) of the HN data set, all loci and popula-tions were included. Sampled localities were pooled into three groups (CI, CN and HN), and six simple, yet competing, introduction scenar-ios were generated under a coalescent framework (Figure 5: 1–6), to focus the computational efforts on probable introduction scenarios rather than an exhaustive list of possibilities (see Appendix 2 for de-tailed introduction scenarios). As the STRUCTURE results revealed that a subsample of the invasive South African M. dolomieu
individu-als (CIS) were more closely related to the historic native samples than
to the remaining SA individuals (CI) (predominantly individuals from
populations BE and OL; Figure 4: b), we simulated nine additional scenarios to test the theory of multiple introductions (Figure 5: A–I; Appendix 2). At last, as suggested by Guillemaud, Beaumont, Ciosi, Cornuet, and Estoup (2010), three supplementary scenarios were
simulated to determine whether the two SA groupings (CI and CIS)
originated from (a) a single serial introduction from the source popu-lation (CN + HN), (b) two independent introduction events from the same source or (c) an unsampled source population (Figure 5: i–iii; Appendix 2). To prevent overparameterization, parameters were specified according to the program guidelines (Cornuet et al., 2014). First, we performed a pre- evaluation of the data set to ensure that at least one scenario and its associated priors could generate simulated data sets similar to that of the observed. This was accomplished by simulating 100,000 data sets and comparing summary statistics for both simulated single- sample (i.e., mean number of alleles, genetic di-versity and allele size variance across loci) and two- sample statistics (i.e., mean genetic diversity, number of alleles, allele size variance, mean index of classification, shared allele distance, distance between
samples and FST) to the observed data (Cornuet et al., 2014). As the
mean M index across loci (Garza & Williamson, 2001) was initially developed with conservation planning in mind, this statistic does not perform well with small, unequal sampling sizes and small starting population sizes (Garza & Williamson, 2001). Hence, it was excluded from the summary statistics used in the current analyses. Next, we
simulated 106 data sets per scenario before calculating the posterior
probability (PP) for each. Scenarios were subsequently compared through a logistic regression, which was conducted on the linear dis-criminant analysis components (Cornuet et al., 2014). Each scenarios error rate was evaluated by generating 100 pseudo- observed data sets, using parameter values obtained from one of the scenarios (e.g., scenario 1). The type I error rate was determined by counting the number of times the PPs were higher for any scenario other than the chosen scenario, divided by the number of pseudo- observed data sets (i.e., 100), while the type II error rate was calculated by counting the number of pseudo- observed data sets that unrightfully received the highest PP support (Cornuet, Ravigne, & Estoup, 2010).
3 | RESULTS
3.1 | Contemporary mtDNA analyses
A total of 292 M. dolomieu specimens collected from eight river systems in the invasive SA range (CI) were successfully sequenced for 306 bp of cytb and 979 bp of CR, while the nine native USA (CN) localities yielded a total of 209 and 174 successfully sequenced
M. dolomieu specimens for cytb and CR, respectively. Both cytb and
CR rendered fewer haplotypes for the CN range when compared to the CI range, but similar haplotype and nucleotide diversity levels were observed (Table 2). Overall, high haplotype and low nucleotide diversity levels were observed for both native (cytb: h = 0.976 ± 0.005, π = 0.051 ± 0.025; CR: h = 0.977 ± 0.007, π = 0.044 ± 0.021) and invasive (cytb: h = 0.967 ± 0.007, π = 0.087 ± 0.043; CR:
T A B LE 2 G en et ic d iv er si ty in di ce s (h ap lo ty pe (h ) a nd n uc le ot id e (π )) an d ne ut ra lit y te st s (T aj im a’ s D a nd F u’ s Fs ) f or th e co nt em po ra ry in va si ve (C I) an d co nt em po ra ry n at iv e (C N ) r an ge s, ba se d on m tD N A c yt b an d C R . S am pl e si ze is d en ot ed b y n, w hi le H re fe rs to th e nu m be r o f h ap lo ty pe s. S ta tis tic al ly s ig ni fic an t r es ul ts (p < 0 .0 5) a re in di ct ed in b ol d Cy to ch ro m e b ( cy tb ) C on tr ol r eg io n ( C R) n H h π D Fs n H h π D Fs C on te m po ra ry in va si ve S A lo ca lit ie s B E 20 16 0. 96 3 ± 0. 03 3 0.0 66 ± 0.0 34 −1 .6 82 −1 .7 58 21 14 0. 86 7 ± 0. 07 4 0. 08 8 ± 0. 04 4 −2 .2 77 6.1 60 B R 42 33 0. 97 6 ± 0. 01 4 0. 06 1 ± 0. 03 1 −1 .2 95 −9 .8 8 43 33 0. 98 1 ± 0. 01 1 0. 03 6 ± 0. 01 8 −2 .0 11 −4 .3 40 BU 47 30 0. 96 5 ± 0. 01 3 0. 06 1 ± 0. 03 1 −2 .0 04 −4 .5 74 47 35 0. 98 4 ± 0. 00 8 0. 02 0 ± 0. 01 0 −2 .5 94 −1 0. 91 8 D O 35 30 0. 98 7 ± 0. 01 2 0. 26 3 ± 0. 12 9 0. 314 −1 .2 95 36 30 0. 97 9 ± 0. 01 6 0. 08 4 ± 0. 04 1 −2 .5 37 0. 32 1 KO 46 24 0. 75 6 ± 0. 07 1 0. 04 4 ± 0. 02 2 −2 .3 10 −2 .777 45 36 0. 98 4 ± 0. 01 0 0. 01 3 ± 0. 00 7 −1 .7 1 −2 1. 92 4 KR 14 9 0. 83 5 ± 0. 10 1 0. 05 0 ± 0. 02 7 −1 .7 68 0. 833 15 15 1. 00 0 ± 0. 02 4 0. 04 6 ± 0. 02 4 −2 .0 47 −2 .6 42 MP 45 37 0. 98 7 ± 0. 00 9 0. 07 1 ± 0. 03 6 −0 .2 57 −1 1. 88 1 45 31 0. 94 2 ± 0. 02 4 0. 06 3 ± 0. 03 1 −2 .6 46 0. 974 O L 43 24 0. 94 7 ± 0. 02 0 0. 03 3 ± 0. 01 7 −2 .0 71 −5 .4 58 40 17 0. 90 6 ± 0. 02 9 0. 04 5 ± 0. 02 2 −1 .6 03 8. 41 7 O ve ra ll 292 176 0. 96 7 ± 0. 00 7 0. 08 7 ± 0. 04 3 −1 .8 99 −2 3. 54 7 292 17 9 0. 98 5 ± 0. 00 3 0. 03 9 ± 0. 01 9 −2 .7 17 −2 3.6 04 C on te m po ra ry n at iv e U SA lo ca lit ie s DE T 7 7 1. 00 0 ± 0. 07 6 0. 14 4 ± 0. 08 3 0. 76 7 −0 .2 26 — — — — — — H U D 20 15 0. 96 8 ± 0. 02 5 0. 05 0 ± 0. 02 6 −2 .1 40 −1 .6 75 17 17 1. 00 0 ± 0. 02 0 0. 13 4 ± 0. 06 8 0. 69 2 −1 .1 45 LO L 20 16 0. 97 4 ± 0. 02 5 0. 04 0 ± 0. 02 1 −1 .9 40 −3 .6 62 20 13 0. 88 4 ± 0. 06 7 0. 00 1 ± 0. 00 1 −1 .1 74 −1 5. 96 8 N IA 48 31 0. 95 7 ± 0. 01 8 0. 03 2 ± 0. 01 7 −2 .4 45 −1 2. 40 3 38 28 0. 97 6 ± 0. 01 4 0. 01 1 ± 0. 00 6 −2 .1 57 −1 3. 58 3 O N EI 30 26 0. 98 9 ± 0. 01 3 0. 02 2 ± 0. 01 2 −1 .5 45 −2 0. 166 18 17 0. 99 4 ± 0. 02 1 0. 08 2 ± 0. 04 2 −2 .38 9 −0 .8 67 O N EO 10 8 0. 95 6 ± 0. 05 9 0. 15 6 ± 0. 08 4 −0 .6 89 2.7 82 10 10 1. 00 0 ± 0. 04 5 0. 01 2 ± 0. 00 7 −1 .5 75 −4 .1 88 SA R 13 12 0. 98 7 ± 0. 03 5 0. 03 0 ± 0. 01 7 −0 .61 5 −4 .47 1 7 7 1. 00 0 ± 0. 07 6 0. 30 1 ± 0. 16 9 −1 .8 06 2. 17 9 STL 47 34 0. 96 6 ± 0. 01 7 0. 03 2 ± 0. 01 7 −0 .82 9 −1 8. 17 8 51 32 0. 94 2 ± 0. 02 3 0. 00 2 ± 0. 00 1 −1 .9 60 −2 8. 46 4 VE S 14 10 0. 92 3 ± 0. 06 0 0. 02 2 ± 0. 01 2 −1 .9 50 −2 .1 14 13 10 0. 96 2 ± 0. 04 1 0. 05 9 ± 0. 03 1 −1 .4 18 2.7 03 O ve ra ll 20 9 11 6 0. 97 6 ± 0. 00 5 0. 05 1 ± 0. 02 5 −2 .1 91 −2 3. 87 0 174 11 7 0. 97 7 ± 0. 00 7 0. 04 4 ± 0. 02 1 −1 .8 29 −2 3. 75 6
between sampling localities and gene fragment (Table 2). In particular, overall nucleotide diversity was higher for cytb in the CI populations (Table 2). Significant deviations from neutrality were observed for Tajima’s D and Fu’s Fs in both native and invasive range and both gene
fragments (Table 2). Pairwise FST measures revealed two significantly
differentiated groupings: CI and CN (Supporting Information Table S2), with comparisons between localities from the two groups ranging
from FST = 0.013 to FST = 0.172 (both p < 0.05) for cytb (DO—SAR and
KO—VES) and FST = 0.013 to FST = 0.125 (both p < 0.05) for CR (KR—
NIA and BE—LOL; Supporting Information Table S2). With regard to the cytb gene fragment, the CN DET population was not significantly different from any of the CI populations, except KO. Similarly, for the CR, the CN populations ONEO and SAR were not significantly different from the majority of CI populations (Supporting Information Table S2). Significant within grouping, differentiation (though markedly less so for the USA cytb) was also observed in both cytb and CR (Supporting Information Table S2). The AMOVA results revealed that the largest proportion of genetic variation (cytb: 94.79%; CR: 95.79%) was distributed within each population, with very little variation observed between the groups (cytb: 2.15%; CR: 1.58%), as well as among populations within groups (cytb: 3.06%; CR: 2.26%). All variance components were significantly different from 0 (p < 0.001).
3.2 | Contemporary and historical
microsatellite analyses
A total of 519 contemporary sampled specimens, representing both invasive (n = 306; eight localities) and native (n = 213; nine localities) ranges, were successfully genotyped for nine microsatellite loci, while 53 museum samples, representing 11 localities within the historical native range, were successfully genotyped for eight microsatellite loci. Neither of the three groups (CI, CN and HN) displayed amplification errors (i.e., large allele dropout, stuttering), nor did any loci exhibit linkage disequilibrium. FreeNA (Chapuis & Estoup, 2007) revealed the presence of null alleles in microsatellite Mdo9 within the HN samples, but this was not the case for either of the contemporary groups. Deviations from HWE were observed in two CI populations (BE and OL) as well as the HN population (FIS:
BE = 0.26, OL = 0.17, MUS = 0.43; Supporting Information Table S3). Further inspection revealed that this deviation was due to a heterozygote deficit within each of the three populations, suggesting the presence of a Wahlund effect (Wahlund, 1928; Waples, 2014), albeit negligible (Guillemaud et al., 2015; Lye, Lepais, & Goulson, 2011). Hence, all further analyses were conducted on the complete data set. The number of alleles (Na) and allelic richness (AR) were consistently higher in the HN data set, and similar between the two contemporary data sets: HN AR = 4.25, CI AR = 1.79–3.15, CN AR = 2.17–2.69 (Supporting Information Table S3). Multilocus
genetic diversity (observed heterozygosity, HO) ranged from 0.39
(ONEI) to 0.59 (DET), while levels of expected heterozygosity (HE)
ranged from 0.35 (MP) to 0.73 (MUS) across all loci.
There was substantial variation in observed heterozygosity
(HO) among populations and loci, with reservoirs (catchment size
<5,000 km2) consistently displaying lower levels of H
O (Figure 2,
Supporting Information Figure S2). Moreover, the ANOVA revealed
significant differences in HO between the three groups (F2,214 = 22.90,
p = <0.001), with HO being higher for HN compared to both con-temporary groups (Bonferroni post hoc test p < 0.001). A significant
marker effect (F7,214 = 19.82, p < 0.001) was, however, observed.
Overall, FST among HN samples was significantly low (FST = 0.013;
p < 0.05), but this was not so for the CI (FST = 0.211; p < 0.05) and CN (FST = 0.091; p < 0.05) populations. Likewise, pairwise FST values revealed significant population differentiation among CI populations, ranging from FST = 0.066–0.469 (DO—KO and BE—MP), with similar results being observed when comparing populations across all three groups, that is, CI, CN and HN (FST = 0.123–0.537; MP—SAR and OL— MUS; Supporting Information Table S4). In contrast, CN populations displayed significantly less population differentiation among sampled localities within this group (FST = 0.072–0.129; LOL—NIA and SAR— STL; Supporting Information Table S4). As predicted, the Wilcoxon rank test revealed a significant excess of heterozygotes for both CI and CN under the IAM model (p = 0.002 and p = 0.010, respectively), but this was not the case under the TPM model (CI: p = 0.230; CN:
p =
0.473). Similarly, no significant excess of heterozygotes was de-tected for the HN population (IAM: p = 0.473; TPM: p = 0.998). The principal component analysis (PCA), based on allelic fre-quencies, revealed two distinct groups along the first two axes: the first comprising both CN and CI populations and the second com-prising the HN populations (Figure 3). Limited genetic associations between the two groups were observed. The Bayesian clustering analyses conducted in STRUCTURE revealed population substruc- turing within the CI localities, with Delta K (Evanno et al., 2005) re-trieving K = 5 as the most probable number of clusters (Figure 4a). Both CI reservoirs (BU and MP) were represented by their own clus-ter and showed very little population variation, corroborating the genetic diversity results (Figure 2; Supporting Information Table S3). The remaining six CI populations, however, displayed substantial lev-els of admixture, in particular localities BE and OL (Figure 4a). The CN populations exhibited high levels of population admixture indic-ative of shallow population differentiation, with Delta K revealing the most probable K = 4 (Figure 4a). Similar levels of admixture and Delta K (K = 4) were obtained for the HN populations (Figure 4a). To determine the most probable source population of the CI popula-tions, all 28 localities were combined (Figure 4b). Delta K revealed the most probable number of clusters to be K = 3, with each cluster representing a group, although admixture between the two contem-porary groups was observed. Interestingly, a subset of individuals within the CI localities BE and OL (and to a lesser extent DO and KO) shared a cluster with HN, but this was not the case for any of the CN populations, despite overlapping sampling localities (DET, HUD, Susquehanna River: LOL, ONEO, VES, SU; Table 1; Figure 4b). The ABC analysis consistently supported the notion of a more recent introduction. The first set of scenarios tested (Scenarios 1–6; Figure 5) revealed that Scenario 2 had the highest posterior proba-bility (Supporting Information Table S5). The second set of analyses (Scenario A–I; Supporting Information Figure S1) supported both
Scenarios C and F (Supporting Information Table S5). The third set of simulations (Scenarios i–iii; Supporting Information Figure S1), where we tested for a single versus multiple introductions from a single source or an unsampled source population, was inconclu- sive. Scenario iii (unsampled source population) did, however, mar-ginally receive the most support (Supporting Information Table S5). Type I and Type II error rates were marginally low for the first two sets of simulations conducted (Supporting Information Table S5), but not for the third simulation (Supporting Information Table S5).
4 | DISCUSSION
Numerous studies have compared genetic diversity levels across native and invasive ranges in an attempt to reconstruct the invasion history of invasive species (reviewed in Dlugosch & Parker, 2008; Lee, Patel, Conlan, Wainwright, & Hipkin, 2004; Novak & Mack, 2005; Rius & Darling, 2014; Roman & Darling, 2007), yet most of these studies only utilize contemporary genetic specimens. This, however, does not account for allele frequency shifts and assumes that the contemporary population structure within the native range would correspond to that of the historically native population. Using
M. dolomieu as a study organism and incorporating both historical
and contemporary native and invaded range samples, our results reveal that genetic diversity and population dynamics can indeed differ across both spatial and temporal scales.
4.1 | Genetic diversity through space and time
Elevated levels of genetic diversity were observed in the contem- porary invasive (CI) range when compared to the contemporary na-tive (CN) range, contradicting the general assumption that genetic diversity is lower in recently invaded ranges than in long- established native populations. However, when comparing all three groups, the historical native (HN) range displayed the highest levels of heterozy-gosity, number of alleles (Na) and allelic richness (AR). Although this might have resulted from a statistical artefact due to the smaller sam-ple size for the HN range, similar findings were previously reported for Atlantic salmon (Salmo salar; Nielsen, Hansen, & Loeschcke, 1997). The authors observed a significant decrease in Na for the con-temporary population when compared to samples collected 60 years before, likely due to a recent genetic bottleneck. Our results sup-port this proposition, as the CN population exhibited high haplotype, but low nucleotide genetic diversity, as well as significantly negative Tajima’s D and Fu’s Fs levels, all of which are commonly observed in a population that had undergone a genetic bottleneck before ex-periencing expansion (Grant & Bowen, 1998). Moreover, the lack of population structure in the CN range, as well as low AR and Na, fur- ther supports this notion. Strong and sustained declines in popula-tion size, such as the ones experienced by commercially exploited species, are known to leave signatures in the genetic diversity of spe-cies, in particular by reducing Na and AR (Pinsky & Palumbi, 2014). Thus, the observed contemporary population dynamics of M.do-lomieu in its native range might have resulted from the interaction between overfishing and restocking events during the last two cen-turies (Long et al., 2015). Micropterus dolomieu has been harvested both commercially and recreationally since the 1800s and has expe- rienced several population declines and even extirpations in some lo-calities (Marsh, 1867). This led the US government to start breeding programmes and enforce stricter regulations on fishing in the 1870s (Long et al., 2015). In 1903 alone, ~500,000 juvenile black bass were released into waterbodies across the USA (Bowers, 1905; Long et al., 2015; Loppnow et al., 2013). Concomitant fluctuations in population sizes are likely to have left genomic signatures and contributed to the observed elevated admixture in CN populations, as reintroduc-tions were conducted without consideration for genetic population F I G U R E 2 A stacked bar graph representing the variation in observed heterozygosity (HO) among populations and loci between the three groups (CI—contemporary invasive SA, CN— contemporary native USA, HN—historical native USA). Reservoirs (excluding Lake Erie (LE)) are indicated with an asterisk (*)
structure in M. dolomieu. Similar findings have been reported for another exploited freshwater species, the brook charr (Silvanus
fon-
tinalis), with individual admixture levels increasing with stocking in-tensity (Lamaze, Sauvage, Marie, Garant, & Bernatchez, 2012; Marie, Bernatchez, & Garant, 2010).
4.2 | Population substructuring in an invaded range
Elevated levels of genetic diversity are, however, not uncommon in invasive species in a novel invaded range and are often attributedto multiple introductions and/or population mixture (see Rius & Darling, 2014 for a comprehensive review). The results from the STRUCTURE analyses appear to contradict the historical records stating that invasive South African M. dolomieu populations originate from a single introductory event from the USA in 1937. A genetic cluster encompassing samples from the Berg (BE: n = 14), Olifants (OL: n = 7), Doring (DO: n = 2), and Kouga (KO: n = 1) Rivers suggests shared ancestry with the HN samples, but the remainder of the in-vasive South African populations belong to four additional clusters, hinting at the idea of multiple introductions. The ABC results sup-port this notion, as the best- fit scenario suggested a second, more recent, introduction from North America (Scenario 2). Furthermore, when considering the invasive South African individuals associated
with the HN STRUCTURE cluster as a separate South African popu-lation (CIS), the ABC analyses supported the STRUCTURE results
and suggested at least two introductions: one coinciding with the recorded historic introduction and at least one more recent intro- duction. Indeed, the observed admixture between CI and CN sug-gests that the more recent introduction also originated from the USA. Unexpectedly, no support was obtained for either scenario examining single versus multiple introductions from a single source (Scenarios i and ii), nor any scenario postulating admixture (Scenarios 4, 5, 6). This may be due to several factors, such as the unequal sam-ple sizes between HN and CI/CN range, the simplicity of the ABC models, or perhaps it could be ascribed to the fact that the HN pop-ulation was not in HWE. Furthermore, the presence of a temporal Wahlund effect within the HN range, likely due to the pooling of multiple populations, may too have decreased the accuracy of the ABC results. Although our results support the notion of multiple introductions, this should be interpreted with caution as several factors may be responsible for this pattern, including an unsampled source population, postinvasion genetic drift, insufficient marker resolution and admixture in the source population (Chown et al., 2015; Gray et al., 2014). Given that hatcheries make use of artificial selection techniques to enhance species production and abundance (e.g., Aprahamian, Smith, McGinnity, McKelvey, & Taylor, 2003; Lamaze et al., 2012), it is possible that the introduced M. dolomieu F I G U R E 3 Principal component analysis (PCA) conducted on the combined microsatellite genotypes for the three groups (i.e., CI—contemporary invasive SA, CN—contemporary native USA, HN—historical native USA). Each dot represents a genotyped individual, and colours correspond to sampled localities. Variance explained in parentheses F I G U R E 4 STRUCTURE plots representing the population structure within (a) each of the three groups (CI—contemporary invasive SA, CN—contemporary native USA, HN—historical native USA) when ran independently, and (b) population structure for all localities combined into a single run. Each genotyped individual is represented by a vertical line, with each lines’ colour proportional to the cluster membership of the individual
were of admixed or hybrid origin, as has been reported for stockings of S. fontinalis (Cooper, Miller, & Kapuscinski, 2010; Lamaze et al., 2012; Sloss, Jennings, Franckowiak, & Pratt, 2008).
Invasive species capable of harbouring large, genetically diverse source populations are thought to make better invaders (Gaither, Bowen, & Toonen, 2013), as they are equipped with higher adaptive potential (Dlugosch, 2006; Lavergne & Molofsky, 2007; Wellband & Heath, 2017). Within the invasive South African range, M. dolomieu experiences an array of climatic conditions with fluctuating rainfall and temperature regimes (Rutherford, Mucina, & Powrie, 2006). However, despite this, M. dolomieu has not only survived, but also established viable populations and spread throughout the systems into which it was introduced (Van Der Walt, Weyl, Woodford, & Radloff, 2016). Although the initial introduced individuals may have been of admixed stock, the substantial admixture observed among M. dolomieu popu-lations in the invaded range may also have resulted from hybridization post introduction (Diedericks, Henriques, von der Heyden, Weyl, & Hui, 2018) as has been observed for M. dolomieu introductions else-where (Avise et al., 1997; Bagley, Mayden, Roe, Holznagel, & Harris, 2011; Pipas & Bulow, 1998; Whitmore & Butler, 1982; Whitmore & Hellier, 1988). Further, although sampling was conducted away from known angling “hotspots,” M. dolomieu are popular angling species and human- mediated, long- distance dispersal via intentional stock-ing cannot be excluded as a mechanism. Such long- distance (human- mediated) dispersal events are known to increase population mixing, ultimately increasing the species’ genetic diversity and hence, adapt-ability to the novel environment (Berthouly- Salazar et al., 2013).
4.3 | The influence of sampling design on
genetic diversity
Molecular techniques are indispensable tools in invasion biology (Blanchet, 2012; Muirhead et al., 2008), particularly for reconstruct-ing species invasion histories and routes (Estoup & Guillemaud, 2010; Guillemaud et al., 2010, 2015; Wilson, Dormontt, Prentis, Lowe, & Richardson, 2009). However, sampling problems such as the num-ber of native versus invasive populations sampled and the number of individuals sampled per population may hinder the accuracy of
the molecular markers to identify the source population (Guillemaud et al., 2010). To date, however, no study has looked at the effect that “sampling locality” may have on each populations’ genetic compo-sition and, hence, genetic diversity. For example, aquatic freshwa-ter species, particularly fish, are often collected from natural lakes or man- made reservoirs due to the ease of collection and the large number of individuals present. These specific sampling sites, how-ever, often display much lower levels of genetic variability when compared to rivers, as suggested by our results (localities BU and MP in the invasive range). Similarly, a recent study reconstructing the invasion history of the largemouth bass, Micropterus salmoides, identified extremely low levels of neutral genetic diversity within in-vasive populations in lentic environments with limited connectivity (Hargrove, Weyl, & Austin, 2017). Their results revealed that all lentic populations had allele frequencies dominated by a single allele, but that a population sampled from Kowie Weir, located at the end of a 580 km2 catchment, was more diverse, suggesting multiple introduc- tion events or hybridization between co- occurring Micropterus spe-cies (Hargrove et al., 2017). Thus, choice of sampling locality and, in particular, the degree of isolation are important considerations when assessing the demographic or invasion history of a species.
4.4 | Management implications
Understanding the introduction history of an invasive species is cru-cial when wanting to decide on a management strategy for the species in question (Prentis et al., 2009). Our results reveal a complex demo-graphic history for M. dolomieu, both within its native USA and inva-sive SA range. With regard to management in the native range, our data support the management recommendations by Brewer and Orth (2015) that stocking should be guided by a rangewide analysis of ge-netic variation. In South Africa, eradication of M. dolomieu is no longer a feasible option due to the magnitude of the invasion, and the current management strategy is to prevent spread into previously uninvaded catchments by restricting stocking (see Woodford et al., 2017). This is a prudent strategy as the facilitation of strategies that might further increase genetic diversity, thought to assist population establishment, persistence and ultimately local adaptation to novel environments,F I G U R E 5 Probable introduction scenarios tested with approximate Bayesian computation as implemented in DIYABC. CI—contemporary invasive SA, CN—contemporary native USA, HN— historical native USA, GH—unsampled ghost population. The arrow indicates time expressed in generations (not to scale), with the present indicated with an asterisk
may increase the fitness of this already highly successful invader. As our study demonstrates the possibility of undocumented M.
dolo-mieu introductions into the country, it is imperative that South Africa
strictly enforces its current legislation with regard to avoiding new introductions of this already invasive species. In addition, introduc-tions even in river systems that have already been invaded may aid in increasing the genetic fitness of these already highly successful in-vaders and could facilitate further spread and exacerbate the already considerable impacts on native biota (Van Der Walt et al., 2016). In conclusion, while studies comparing contemporary genetic vari-ation among native and invasive ranges are valuable (Lozier & Cameron, 2009), incorporating historical DNA is essential for monitoring tempo- ral changes in genetic diversity that are often overlooked in compari-sons using only contemporary data (Hansen, 2002; Lozier & Cameron, 2009). Using the smallmouth bass, M. dolomieu, as study organism, our results corroborate the idea that genetic variation can indeed change over spatiotemporal scales. Both CI and CN range displayed high levels of admixture and limited population structuring. Although this pattern is not uncommon for invasive species that have been introduced mul-tiple times, our results suggest that various factors may have played a role in shaping the genetic diversity of the CI range.
Our study highlights the importance of including historical DNA; however, caution should be taken when working with histor- ical specimens as the degraded nature of the DNA not only ham-pers the successful amplification of the specimens (Sefc, Payne, & Sorenson, 2003; Sefc et al., 2007), but also renders it susceptible to genotyping discrepancies. Despite this, we recommend that fu- ture studies attempting to infer the demographic history of inva-sive species should incorporate native historical samples. ACKNOWLEDGEMENTS The authors would like to thank Louis Bernatchez and two anony-mous reviewers for valuable comments and suggestions on earlier versions of the manuscript. We would like to thank the following peo-ple and institutions for their sample contributions: USA—Erling Holm (Royal Ontario Museum), Douglas Nelson (University of Michigan Museum of Zoology), Jeff Williams (National Museum of Natural History), Mark Sabaj Pérez (Academy of Natural Sciences), Mark Kibbey (Ohio State University Museum), Jason Barnucz (Fisheries and Oceans Canada), Wil Wegman (Ontario Ministry of Natural Resources and Forestry, Aurora District), Gene Gilliland (B.A.S.S. conservation director), Rich Carter (Ohio Department of Natural Resources), Randy Jackson (Department of Natural Resources, Cornell University) and Jeff Loukmas (New York State Department of Environmental Conservation); SA—Chris Broeckhoven, Riaan, Beverley & Adriaan Diedericks, Jacques Aproskie and Craig Fraser (South African Bass Angling Association). A special word of thanks to Francois Roux (Mpumalanga Parks Board) for assistance in obtain-ing sampling permits, anglers and coordinating the sampling trips to the Blyde River Canyon, a world heritage site. We acknowledge use of specimens, infrastructure and equipment provided by the NRF- SAIAB Research and Collections Platforms.
This work is based on research supported partly by the Department of Science and Technology (DST) and National Research Foundation (NRF) of South Africa (Grant Nos. OLFW: 110507, 109015; CH: 89967, 109244) and the DST- NRF Centre of Excellence for Invasion Biology.
CONFLIC T OF INTEREST
None declared.
DATA ARCHIVING STATEMENT
Data available from the Dryad Digital Repository: https://doi. org/10.5061/dryad.5jf41k5.
ORCID
Genevieve Diedericks http://orcid.org/0000-0001-7700-8906
REFERENCES
Aprahamian, M. W., Smith, K. M., McGinnity, P., McKelvey, S., & Taylor, J. (2003). Restocking of salmonids—opportunities and limita-tions. Fisheries Research, 62, 211–227. https://doi.org/10.1016/ S0165-7836(02)00163-7
Avise, J. C., Pierce, P. C., Van Den Avyle, M. J., Smith, M. H., Nelson, W. S., & Asmussen, M. A. (1997). Cytonuclear introgressive swamping and species turnover of bass after an introduction. Journal of Heredity,
88, 14–20. https://doi.org/10.1093/oxfordjournals.jhered.a023049
Bagley, J. C., Mayden, R. L., Roe, K. J., Holznagel, W., & Harris, P. M. (2011). Congeneric phylogeographical sampling reveals poly-phyly and novel biodiversity within black basses (Centrarchidae:
Micropterus). Biological Journal of the Linnean Society, 104, 346–363.
https://doi.org/10.1111/j.1095-8312.2011.01720.x
Baker, H. G., & Stebbins, G. L. (Eds.). (1965). The genetics of colonizing
spe-cies. New York, NY: Academic Press.
Barthel, B. L., Cooke, S. J., Svec, J. H., Suski, C. D., Bunt, C. M., Phelan, F. J. S., … Philipp, D. P. (2008). Divergent life histories among smallmouth bass Micropterus dolomieu inhabiting a connected river—lake system. Journal of Fish Biology, 73, 829–852. https://doi. org/10.1111/j.1095-8649.2008.01972.x
Beneteau, C. L., Walter, R. P., Mandrak, N. E., & Heath, D. D. (2012). Range expansion by invasion: Genetic characterization of invasion of the greenside darted (Etheostoma blennioides) at the northern edge of its distribution. Biological Invasions, 14, 191–201. https://doi. org/10.1007/s10530-011-9996-8
Bernatchez, L., & Danzmann, R. G. (1993). Congruence in control- region sequence and restriction- site variation in mitochondrial DNA of brook charr (Salvelinus fontinalis Mitchill). Molecular Biology and Evolution, 10, 1002–1014.
Berthouly-Salazar, C., Hui, C., Blackburn, T. M., Gaboriaud, C., van Rensburg, B. J., van Vuuren, B. J., … Le Roux, J. J. (2013). Long- distance dispersal maximizes evolutionary potential during rapid geographic range expansion. Molecular Ecology, 22, 5793–5804. https://doi.org/10.1111/mec.12538
Blanchet, S. (2012). The use of molecular tools in invasion biology: An em-phasis on freshwater ecosystems. Fisheries Management and Ecology,
19, 120–132. https://doi.org/10.1111/j.1365-2400.2011.00832.x
Bouzat, J. L. (2000). The importance of control populations for the iden-tification and management of genetic diversity. Genetica, 110, 109– 115. https://doi.org/10.1023/A:1017985522650
